CSSE4004 - Lecture 2

A server which is aware of which clients have a file cached is stateful

Servers which obtain their state from cookies are stateful

Stateful servers makes it more difficult to handle failures

A stateless server can change its own state without informing clients

Because threads share the address space and therefore have lower overhead to start

To provide total isolation between different tasks within a process

Because processes can only be used for weak consistency

Because threads do not require state saving when interrupted

Interpreted languages running on a virtual machine

Passing the code by reference

Dynamically reconfiguring the target system

Using a cloud computing platform instead

To change flow of control and allow additional code to be executed

To block unauthorized or excessive RPC traffic

To update the code with new versions

To support binary encoding

Abstraction of activity within process

An isolated task containing an independent address space

An abstract task that can be passed between nodes within a distributed system over a network

The owner of a process

Hide network latency by reacting to next request while previous being transmitted

By migrating the thread to the client machine which is more powerful

Because threaded servers perform better with threaded clients

It doesn't, threads are vulnerable to blocking execution

Reduce communication

Continue when disconnected

Dynamically deploy components

Because .NET is only supported on Windows platforms

Because I do not trust the code and wish to run it on an isolated system

Because the product of the code is larger than the code itself

Because the client is much more powerful than the server

Because the same code will be used many times over and over time the data produced by the code will exceed the size of the code

Process, thread, communication and memory management

Process, thread, reliability and memory management

Security, IPC, threading and recovery

Concurrency, resource isolation, communication and memory management

Prevents simple scaling to multiprocessor

Can only serve multiple concurrent clients if those clients are threaded

Generally perform faster by avoiding context switching

Does not scale up to serving many clients